Functional traits of medicinal plant species under different ecological conditions

Medicinal plant species comprise 21% of the Slovakian native flora. Thirteen species, we focused on, grew at a control site (botanical garden) and three natural sites with different ecological conditions. We described effects of irradiance and soil organic matter content on leaf functional traits, such are specific leaf area and leaf dry matter content. For the formation of plant functional groups according to different leaf characteristics, the metric multidimensional scaling (MDS) ordination diagram was applied. Photosynthetically active radiation (PAR) had a more significant effect on studied leaf traits than content of soil organic matter (SOM). Moreover, we studied plasticity of these leaf functional traits and formed plant functional groups based on them. Urtica dioica and Geum urbanum possessed the greatest plasticity of all studied species.


Introduction
There is growing recognition that classifying terrestrial plant species based on their function rather than their taxonomic identity is a promising way forward to tackling important ecological questions at the scale of ecosystems, landscapes or biomes (e.g., Cornelissen et al. 2003). These questions include vegetation responses to and vegetation effects on environmental change, such as climatic change, atmospheric chemistry or land use. To provide mechanistic insight into the rules governing plant species coexistence and diversity, plant community ecologists are increasingly quantifying functional trait values for species growing in a range of communities.
Plant functional traits are the features (morphological, physiological, phenological) that represent ecological strategies and determine how plants respond to environmental factors (Amaral et al. 2021), affect other trophic levels and influence ecosystem properties (Pérez-Harguindeguy et al. 2013). Variation in plant functional traits results from evolutionary and environmental drivers that operate at multiple scales, making it challenging to distinguish among them (Reich et al. 2003). Since morphological traits, such as specific leaf area (SLA) and leaf dry matter content (LDMC), reflect the leaf economic spectrum and plant adaptation (Wright et al. 2004) as well as they are used to classified ecological strategies according to CSR theory (Grime 2001).
From studied ecological factors, we focused on the photosynthetically active radiation (PAR) and soil organic matter (SOM). The level of irradiance is an important ecological factor on which all photo-autotrophic plants depend. Low light intensities pose stresses on plants because irradiance limits photosynthesis and thus net carbon gain and plant growth (Gitelson et al. 2021). On the other hand, high irradiances that are usually associated with high doses of ultraviolet radiation significantly reduce the photosynthetic activity of shade acclimated plants (Klem et al. 2012). For a single leaf, SLA is negatively correlated to light intensity (Ratjen and Kage 2013). Lambers and Oliveira (2019) stressed that plants growing in shady conditions invest relatively more of the products of photosynthesis and other resources in leaf area. Their leaves are relatively thin and have a high SLA and low leaf mass density. These results correspond with the findings of Cornelissen et al. (2003) that species growing in the herb layer of a forest are shade-tolerant with high values of SLA. Shipley and Meziane (1998) have shown that both mineral nutrition and irradiance can have complex and interactive effects on SLA. According to Debuis et al. (2013), soil properties have been widely shown to influence plant growth and distribution. Species from highly productive habitats had higher SLA than those from sites of low productivity, although individual species sometimes deviated substantially from the general trend (Poorter and de Jong 1999). Content of the soil organic carbon is an essential characteristic that affect soil fertility and texture with its complex and heterogenous structures although it occupies a minor percentage of the soil weight (Khaled and Fawy 2011).
Phenotypic plasticity can be characterised as a possibility of a single genotype to produce different phenotypes in diverse environments (Sultan 2000). The ability of an organism to change its phenotype in response to diverse environments is an essential characteristic for sessile plants to acclimate to rapid changes in their environment (Stotz et al. 2021).
Functional groups serve as a strategy for representing plant communities and their relationships with the biotic and abiotic environment, without the need for data input for each species (Rogers et al. 2019). In community ecology, plant functional groups are widely used to describe trait variation within and across plant communities (Thomas et al. 2019). The Grime´s CSR (Grime 2001) model constitutes one of the most established systems for plant functional types. The primary strategies (competitive ability, adaptation to stress and adaptation to disturbance) relate to level of disturbance and the productivity at a given site.
Medicinal plants synthesize many biologically active chemical compounds (secondary metabolites) (Li et al. 2020), that are important for defence against insects, fungi, diseases, and herbivorous animals. Moreover, in a natural plant community, they markedly affect other species through allelopathy. Wild medicinal plants are thus not only a substantial component of natural plant community, but also responsible for biodiversity and stability of natural ecosystems.
Based on the importance of medicinal plant species for natural plant community, the aim of our research was: (i) to identify the trait plasticity using two functional traits (SLA and LDMC) for all 13 studied species; (ii) to classify species into functional groups based on their leaf characteristics within a community; (iii) to characterize the species from the perspective of Grime´s plant strategies; (iv) to discuss the relationship between PAR or soil organic matter content and leaf characteristics.

Research site characteristics
Four research sites in the region of the city Bratislava (SW Slovakia) were chosen. The first research site (BG), the Garden of Medicinal Plants, Faculty of Pharmacy, Comenius University in Bratislava (BG) (GPS coordinates: 48.1419761 N;17.1894686E), served as a control site. Approx. 850 medicinal plant species or potentially medicinal plants are grown in the garden, and the plants are used for both research and education. The garden represents a site with regular management (cultivation, fertilization, and irrigation). Thirteen medicinal plant species from this site were used: Aegopodium podagraria L., Fragaria vesca L., Galium odoratum (L.) Scop., Geum urbanum L., Glechoma hederacea L., Hedera helix L., Hypericum perforatum L., Impatiens glandulifera Royle, Plantago lanceolata L., Prunella vulgaris L., Solidago gigantea Aiton, Tussilago farfara L., and Urtica dioica L. The studied medicinal species were chosen based on previous experiences and because they occurred in the control site (BG) and in another experimental site at the same time.
The third experimental site was an alluvial forest (PB) in the Bratislava district Podunajské Biskupice (GPS coordinates: 48.0799625 N;17.2022244E). This forest represented a more shaded site than the previous two sites with a higher soil organic matter content. Five plant species Galium odoratum, Geum urbanum, Hedera helix, Glechoma hederacea, and Urtica dioica, were sampled.
The final experimental research site was an anthropogenically influenced biotope (ST) close to the Faculty of Electrical engineering and Information technology STU in Bratislava (GPS coordinates: 48.1530903 N; 17.0723700E). This site was without regular irrigation, with occasional mowing, but had the lowest soil organic matter. Samples were taken of Hypericum perforatum and Plantago lanceolata.
In (BG), all 13 species were sampled. In each of (R), (PB), and (ST), a subset of the species was sampled, wereby these subsets were not overlapping.

Indication of the weather
Considering the importance of ecological effects on plants and vegetation, the summer period (June -August) 2018 in Slovakia could be characterized as relatively warm, compared with long-term records (Anonymus 2022): the mean daily temperature was 21.9 °C. In the warm half-year (April -September) 2018 a record number (135) of summer days (days when maximal air temperature ≥ 25 °C) was registered (Faško and Pecho 2018). During the growing season (March -September) 2018, sunshine duration was 1787 h, average temperature was 17.8 °C, and precipitation reached 427.4 mm (hydrometeorological station of Slovak Hydrometeorological Institute, Bratislava -airport). Similarly, the summer period (June -August) 2019 was relatively warm, compared with long term records (Anonymous 2022), with the highest temperature of 37.6 °C (Beránek and Faško 2020). During the growing season (March -September) 2019, the sunshine duration was 1646 h, average temperature was 17.4 °C, and precipitation reached 301.8 mm (hydrometeorological station Bratislava -airport). Presented data of the sunshine duration, temperature, and precipitation were provided by the Slovak Hydrometeorological Institute in Bratislava (Anonymus 2021(Anonymus , 2022.

Fieldwork
The research was conducted in 2018 and 2019 (June and July). For the measurement of leaf functional traits we used 10 samples (from each species and from each site) fully expanded physiologicaly mature leaves ("representative leaf method", Cicák 1998). On each above-mentioned site, PAR was measured (μmol m −2 s −1 , MQS Cosine Corrected Mini Quantum Sensor, ICT International, Armidale, NSW, Australia) under clear sky conditions (anticyclone weather condition) at the level of sampled leaves. PAR was determined when the samples of the study plants were taken -at least 5 times (in experimental site (PB) PAR was measured 30 times (15 times in shade and 15 times on sunfleck). The low PAR values mean that plants (for example Galium odoratum on BG site) were permanently grown during growing season in the shade of the trees. Three soil samples were taken at a 0 − 10 cm depth on each site.

Plant samples processing
The measurement of the functional traits of the plants was in accordance with the procedure described in Cornelissen et al. (2003). The leaf characteristics of the plants that have been To determine SLA and LDMC values, the following formulas were applied:

Soil sample processing
Soil samples were air dried and passed through a 2-mm sieve prior to analys. Soil organic carbon (SOC) was determined by potassium dichromate oxidation (Nelson and Sommers 1996). The percentage of soil organic matter (SOM) was calculated from SOC using the conversion factor 1.724 (Nair 1993). Soil texture was determined by the pipette method (Gee and Bauder 2002) and the soil textural class was found using the USDA textural triangle (USDA 1987). Soil pH was measured in a 1:2.5 soil:0.01 mol CaCl 2 suspension.

Data analysis
The maximum, minimum and mean PAR values are given for control and experimental sites. Mean values and standard deviations were calculated for soil properties at each site. Oneway ANOVA and Tukey's post-hoc test was used to assess significant differences in SOM contents between the sites.
For both analysed leaf traits (SLA and LDMC), the mean values and bootstrapped 95% confidence intervals were determined for each plant species at the control and experimental site. To calculate the bootstrap confidence intervals (bootstrap replicates = 1000), the "boot" package was used (Canty and Ripley 2021).
A Gower dissimilarity matrix based on leaf trait values of each species was used in the metric multidimensional scaling (MDS). The MDS was carried out to identify the position of species centroids at each sampling site and to determine the shift of species centroids between control and experimental site, as well as to evaluate leaf trait plasticity for each species (bivariate ellipse − standard deviation + 95% confidence limits). To the created ordination diagram, PAR and SOM were projected as supplementary variables.
The same Gower dissimilarity was used in the MDS to determine species groups based on the analysed leaf traits. Generalized additive models with a thin plate spline were used to predict and plot the surface of the PAR and SOM gradient on the MDS ordination diagram. In the MDS plot, values of classification trait criteria were obtained from classification and regression tree analyses (CART). The result of the CART was projected to the ordination diagram through the orthogonal projection of each trait classification criterion on the trait vector in the ordination. In the CART model, the relevant number of splits was controlled by the complexity parameter (cp), which imposes a penalty to the tree for having too many splits (i.e., if the next splitting does not significantly improve the overall quality of the previous model). The cp value was set to 0.01. The MDS and CART were performed in the "Vegan 2.5-7" package (Oksanen et al. 2020) and "rpart 4.1-15" package (Therneau and Atkinson 2019), respectively. All used packages were run under the R 3.6.3 (R Core Team 2020) software environment.

Characteristics of the study sites
The values of PAR and soil properties are given in Tables 1 and  2, respectively. The changes in PAR values can be identified with    the gradient of the first as well as second ordination axis (Fig. 1c). Among the soil properties, only SOM showed a significant effect. This is related to the large range of SOM contents and the differences between study sites. The differences were significant between soil samples of Hedera helix, Galium odoratum, Geum urbanum at (BG) and (PB) sites (Tukey's p < 0.01), and between Glechoma hederacea at (BG) and (PB) sites (p < 0.01). Soil samples from the (PB) site had significantly higher SOM contents compared to (R) and (ST) sites (p < 0.01), except for Impatiens glandulifera soil at the (R) site (p = 0.061). A significantly higher SOM content was also recorded in Impatiens glandulifera soil at the (R) site compared to (ST) site (p < 0.01). The highest value of SOM (21.0%) was determined in the soil of Urtica dioica (BG) ( Table 2). Other differences in SOM between experimental sites were not statistically significant. The changes in SOM content were reflected especially in the gradient of the second ordination axis. Based on the result of the multiple regression, PAR explains five times more information in the first two ordination axes than SOM content (Table 3).

Specific leaf area and leaf dry matter content
For the studied species, mean values of SLA ranged from 10.3 m 2 kg −1 to 66.3 m 2 kg −1 (Fig. 2a). The mean values of LDMC were between 0.12 kg kg −1 and 0.38 kg kg −1 (Fig. 2b). According to the overlap of the bootstrapped 95% Fig. 1 Results of the MDS showing the species plasticity in two studied functional traits: SLA -specific leaf area, LDMC -leaf dry matter content. Explanation: The arrows represent the shift of centroid for the particular species between the control and experimental sites (the beginning of the arrow marks the centroid for control site and the end of the arrow marks the centroid for experimental site). PAR -photosynthetically active radiation, SOM -soil organic matter. For explanation of plant species and study sites, see Table 1 Table 3 Table 1 confidence interval of the ordered mean values, six groups were identified for SLA, and four groups for LDMC.

Leaf trait plasticity and classification of functional groups
For the evaluation of species plasticity in analysed leaf traits, MDS was applied (Fig. 1). The first two axes of the MDS represent 85.6% of the total variability in trait values of analysed species. The changes in LDMC were more or less identical with the gradient of the first ordination axis, whereas the changes in SLA related to the gradient of the first as well as the second ordination axis (Fig. 1b). The area of the prediction ellipses reflecting the species plasticity in two analyzed traits ranged from 0.05 to 1.03 (Galium odoratum − 0.05, Impatiens glandulifera − 0.06, Solidago gigantea − 0.07, Plantago lanceolata − 0.08, Glechoma hederacea − 0.1, Aegopodium podagraria − 0.14, Hedera helix − 0.14, Hypericum perforatum − 0.17, Fragaria vesca − 0.2, Prunella vulgaris − 0.23, Tussilago farfara − 0.34, Geum urbanum − 0.38, and Urtica dioica − 1.03) (Fig. 1a). Through the projection of the results of CART analysis to the MDS, we identified seven functional groups based on the leaf characteristics (Fig. 3). The classification on the highest level was based on the leaf characteristic LDMC < 0.23 kg kg −1 (LDMC of the species on the left side of the plot is < 0.23 kg kg −1 and LDMC of the species on the right side of the plot is ≥ 0.23 kg kg −1 ). Classification at the second hierarchical level was determined by SLA values on both sides of the diagram. On the left side (LDMC < 0.23 kg kg −1 ), species were divided by the classification criterion of the SLA < 19.3 m 2 kg −1 . The species (Plantago lanceolata (BG), and Tussilago farfara (BG)) with SLA < 19.3 m 2 kg −1 grew in the habitat with PAR higher than 1600 μmol m −2 s −1 and SOM content lower Prunella vulgaris (R), Solidago gigantea (BG), S. gigantea (R), Hedera helix (PB), Hypericum perforatum (BG), Urtica dioica (BG), and Hypericum perforatum (ST) belonged to the group of species with LDMC ≥ 0.23 kg kg −1 and with 19.1 m 2 kg −1 ≤ SLA ≤ 29.6 m 2 kg −1 . Species (Prunella vulgaris (BG), Geum urbanum (BG), Aegopodium podagraria (BG), Hedera helix (BG), and Fragaria vesca (BG)) with LDMC ≥ 0.23 kg kg −1 and with SLA < 19.1 m 2 kg −1 grew in the environment with PAR higher than 1600 μmol m −2 s −1 and SOM content less than 5%. Subsequently, these species were divided into two groups based on criterion LDMC ˃ 0.34 kg kg −1 . Species Hedera helix (BG), and Fragaria vesca (BG) had LDMC higher than 0.34 kg kg −1 .
On the right side of the diagram (LDMC ≥ 0.23 kg kg −1 ), classification at the second hierarchical level was based on SLA < 19.1 m 2 kg −1 . In species with SLA ≥ 19.1 m 2 kg −1 classification at the third hierarchical level were conducted based on SLA ˃ 29.6 m 2 kg −1 . Only one species (Fragaria vesca (R)) had SLA ˃ 29.6 m 2 kg −1 and LDMC ≥ 0.23 kg kg −1 and it grew in the habitat where PAR was less than 400 μmol m −2 s −1 and SOM content was greater than 7%.
High leaf SLA represents a resource-acquisitive plant strategy, while a high leaf dry matter content (LDMC) represents a resource-conservative strategy (Lambers and Poorter 1992). Species with low LDMC tend to be associated with productive, often highly disturbed environments.
Results published by Majeková et al. (2014) confirmed the connection between plant functional traits and population temporal stability, whereby population temporal stability, measured as a coefficient of variation of species' biomass over time, was related to plant traits (including SLA and LDMC) covering different functional trade-offs. Plant functional traits linked to the leaf economic spectrum are important predictors of population stability regardless of both the abiotic and biotic conditions in which plants grew and species phylogenetic relatedness. High values of LDMC are associated with greater temporal stability, indicating that slow-growing species with more conservative economics are generally more stable over time.
Wild medicinal plant species as a substantial component of natural plant communities contribute to the biodiversity and stability of natural ecosystems. Since medicinal plants are frequently exposed to various environmental stresses in their natural conditions, they have evolved physiological, biochemical, and molecular mechanisms to respond to harmful effects of these stresses (Masarovičová et al. 2019). The present results of SLA and LDMC (Figs. 1, 2, and 3) characterized studied medicinal plants from the perspective of leaf functional traits that contribute to our understanding of their position in natural plant communities.
From the meteorological perspective, sunny and warm weather was favourable for photosynthesis and growth of the studied medicinal plant species grown in both the botanical garden and natural communities. However, sunny and warm weather is usually accompanied by drought, especially under natural conditions, like in our experiments.
Plant communities differ in light, nutrients, and water availability, which are important factors in the selection and differentiation of which leaf traits should be used as an indicator of change in the environment (Amaral et al. 2021). Based on the relationships between SLA and LDMC we focused on the two ecological factors: photosynthetically active radiation (PAR) and soil organic matter (SOM). SLA values increase with decreasing PAR and nutrient availability, which have been confirmed by the results of our research. There is a strong link between these two abiotic factors, while leaf blade thickness and mesophyll thickness increase with increasing PAR and nutrient availability without interaction. PAR is a vital source of energy for plants, and plants compete for this source, especially in dense communities. Plants have a variety of photosensory receptors through which they can detect the presence of competing species and subsequently adapt their growth and development strategies (Fiorucci and Fankhauser 2017). The availability of solar energy is a major ecological factor determining the convergence of leaf characteristics in the plant community, with no apparent effect of soil moisture on leaf characteristics (at the stand level), despite the importance of water in the drought-prone ecosystems (Ackerly 2003). The content of organic matter is an important parameter that indicates the overall quality of the soil, as well. It is influenced by many factors (vegetation, climatic conditions, soil type) (Wan et al. 2019).
The ordination diagram (Fig. 2a) shows that in two functional groups with the lowest SLA values includes species growing on the control site (BG) under the highest PAR values (1927 μmol m −2 s −1 ) except Hedera helix that occurred at 1178 μmol m −2 s −1 PAR. Ratjen and Kage (2013), for a single leaf, found that SLA is negatively correlated with light intensity as shown in our results.
Species with the medium values of SLA were assigned to the third and fourth functional group (Fig. 2a) and occurred at sites (BG), (R), (ST), and (PB). According to Lambers and Poorter (1992), these species would have intermediate potential relative growth rates.
The fifth and sixth functional groups presented species with the highest SLA values (Fig. 2a) that occurred mainly on site (PB). These results correspond with the findings of Cornelissen et al. (2003) that species growing in the herb layer of a forest are shade-tolerant with high values of SLA. These species also have a high potential relative growth rate. To these functional groups belong, in addition to species from site (PB), two species from site (BG) -Galium odoratum and Glechoma hederacea. These species were grown under a lower PAR level than the other species at the (BG) site. Aegopodium podagraria occurred on the (R) site with the lowest measured PAR value (2.7 μmol m −2 s −1 (11:50 h), which was manifested in relatively high SLA values (Fig. 2a). Impatiens glandulifera plants grown at site (R) with a mean value of PAR 366 μmol m −2 s −1 .
Species from site (BG) had, in general, lower values of SLA in comparison with plants from other sites. The highest values of this parameter were shown by Glechoma hederacea (PB), indicating the best adaptation of this species to the low irradiance level. This finding agrees with results in Lambers and Oliveira (2019) that plants growing in shade invest relatively more resources into LA, and these leaves are thin with high SLA values.
We divided the studied species into the functional groups according to the LDMC (Fig. 2b) parameters and their 95% confidence intervals in accordance with bootstrapping values.
In the first functional group were assigned species with the lowest LDMC values (Fig. 2b). These species would possess the highest potential relative growth rate of all studied species. LDMC is related to the average density of the leaf tissues (it is also related to leaf thickness) (Smart et al. 2017) and tends to scale with 1/SLA. It has been shown that LDMC correlates negatively with potential relative growth rate and positively with leaf life-span and net primary production of aboveground biomass, but the strengths of these relationships are usually weaker than those involving SLA (because 1/SLA combines leaf tickness and leaf density) (Lambers and Poorter 1992;Smart et al. 2017). This finding was not confirmed for the first functional group because to this group belong four species with summer leaves -Plantago lanceolata, Tussilago farfara, Impatiens glandulifera, and Aegopodium podagraria and only two species with evergreen leaves -Glechoma hederacea, and Galium odoratum. Species with low LDMC tend to be associated with productive, often disturbed environments.
The third functional group included species with summer leaves (Aegopodium podagraria, Hypericum perforatum, Solidago gigantea, and Urtica dioica), and four species with evergreen leaves (Prunella vulgaris, Hedera helix, Fragaria vesca, and Geum urbanum). According to McIntyre (2008), a conservative strategy of plant species for low-productive undisturbed habitats relates to low SLA and high LDMC in contrast to fertile disturbed habitats, which select for high SLA and low LDMC. Leaf characteristics are useful in quantifying the links between vegetation change and ecosystem function that will be a vital part of ecosystem value assessments.
The last functional group included two species: Hedera helix, and Fragaria vesca, having the highest values of LDMC (Fig. 2b). Leaves with high LDMC tend to be relatively tough (Cornelissen et al. 2003;Smart et al. 2017) and are thus assumed to be more resistant to physical hazards (e.g., herbivores, wind, hail) than leaves with low LDMC. Additionally, leaves of Hedera helix are scleromorphic, and leaves of Fragaria vesca are mesomorphic. Since leaves of these species are evergreen, LDMC showed a positive correlation with the lifespan of the leaves.
Since seasonal plasticity may enable plants to cope with adverse environmental conditions (Stotz et al. 2021) and/ or resource variability (Wang et al. 2019), we studied the plasticity of the plants using MDS (Fig. 1). More plastic species (e.g., Urtica dioica, and Geum urbanum) might show a greater ability for adaptation to ecological conditions. Similarly, Galium odoratum, Impatiens glandulifera, and Solidago gigantea are plastic species under given environmental conditions.
We performed a more complex differentiation of the examined species, when we used the metric multidimensional scaling (MDS) (Fig. 3). The first functional group includes Plantago lanceolata (BG), and Tussilago farfara (BG). The species of this first functional group belong to C-S-R plant strategists. The values for individual taxa were modified and extended for the Czech flora by Chytrý et al. (2018Chytrý et al. ( , 2021.

Conclusions
We described effects of irradiance and soil organic matter content on leaf functional traits, such are SLA and LDMC. The applied MDS method showed that PAR had a greater effect on the studied leaf traits than soil organic matter content. Moreover, we studied plasticity of these leaf functional traits and formed plant functional groups based on them. Urtica dioica and Geum urbanum possessed the greatest plasticity of all studied species.
Since plant species of a given functional group responded similarly to environmental variation, we expect them to respond similarly to ecosystem processes and habitat properties. Based on the functional groups we predict the response of the studied medicinal plant species to different ecological conditions. Traits and attributes of plants in plant communities are the outcome of adaptation of species to the environment and acclimation to environmental conditions changing in space and time. Above mentioned approach allowed us to understand and predict the behaviour of individuals and plant species populations in communities and functional interpretation and organisation of plant communities. We emphasize that the main mechanisms by which biodiversity affects stability of ecosystem functions all act through functional traits of organisms that form local communities. Therefore, identification of plant functional groups in a comunity can provide a better understanding and functional comparison of several communities than the 1 3 classical approach based on taxonomy. Since data concerning the functional traits of medicinal plants are scarce, the present results will contribute to this field.